2024 IEEE 5th Women in Technology Conference (WINTECHCON) Automated Camera Quality Calibration & Matching

In semiconductor fabrication processes, often several camera devices are employed to collect data for activities such as surveillance, monitoring, image processing and model training for AI algorithms. Standardized image quality across camera devices is thus crucial for many of these use cases. However, managing multiple camera settings manually is challenging and prone to errors. Hence, an automated solution is required to calibrate all cameras with similar image quality settings in a single operation. In this paper, we develop a novel algorithm to standardize image quality across multiple target camera devices, utilizing optimization techniques and classical machine learning algorithms to minimize differences in camera attributes. To quantify the image similarities between reference and target cameras, we devise a Tool-to-tool matching (TTTM) score. The proposed approach is integrated into an application that enables users to perform single-click image quality matching operations without the need for manual adjustments. Though the proposed solution has been demonstrated and deployed for a semiconductor industrial setup, it is extendable across domains. The solution significantly improves time-efficiency, reducing device tuning time from order of hours to seconds.